61 comments

[ 3.4 ms ] story [ 29.3 ms ] thread
This is such a venerable and ancient class of bugs, going at least as far back as AIX 3. Glad to see they're still makin' 'em like they used to.

(If you had SSH access to a host in your Tailscale ACL, you could log in as `-i` and get a root login.)

you can also add parameters to env vars in some popular cloud providers for the same effect.
I'm somewhat alarmed that the context that this bug was running in was capable of root login. Is there a reason that an SSH login process would, by default, have enough capabilities to facilitate direct root login?
If it runs as your user and can only log in as you, then I wouldn't expect it to be able to become root. But if it can log you in as different users, I would expect that 1. it needs to run from root, and 2. it can log in as root.
We did Tailscale-like SSH reverse tunnels at scale first in 2013 and the main issue has always been that there are no good libraries. Bash scripting around the OpenSSH binaries is pretty much the only way to go.

There's Paramiko, but Python is still a huge liability in memory-constrained systems.

libssh, libssh2 (these are totally independent and unrelated code bases, libssh is maintained by Red Hat for ansible, libssh2 was created for curl)
So, giving access via tailscale but using OpenSSH is safe, right?
as much as handing control to a remote third part is, yes.
Good point. I self host headscale but it also has the ssh feature, probably also insecure.
Not necessarily, it's a clean room implementation. Even if leading dashes was known/documented/tested to implement they might have done it differently. And maybe it was an implementation detail that it was ever allowed, but that's a weird username, headscale implementation happened not to allow it, and nobody ever noticed the discrepancy.
> "Tailscale SSH now rejects usernames with leading dashes."

Really? That's the fix?

A proper fix is to use "--" to separate arguments.

A proper fix is not to shell out to a command at all; use getpwnam(3) or similar.
[delayed]
Are there any actual systems that can run Tailscale and that have faulty getent?
I’m a heavy Tailscale user, so I do trust them quite a bit, but I never used the Tailscale SSH feature. I feel like OpenSSH’s security record is pretty unbeatable, not sure why I’d swap over for such a security-sensitive tool.
I've used it before to access my tailnet machines through a browser on a machine I can't download software on.
I just don't use stranger's machines to access my personal stuff. Possibly compromised stranger's machines. I don't see the benefit about that, as I have more laptops than I need.
Convenience for the most part but in general, I agree. I like having it as an option.
Yeah pretty much just use tailscale as a vpn.. do one thing as they say.
I used it for a bunch of remote monitor boxes to have a way of centrally managing ssh access to things that were often on- and off-line. It was simple and convenient and access was easily revocable.
The SSH vulnerability here only applies if the attacker is already on the network. It violates your Tailscale ACLs, but it's not arbitrary external root ssh access. Arguably that's a more secure starting point than vanilla ssh to publicly accessible machine.
OTOH, if you run vanilla ssh on a publicly accessible machine where only port 22 is open, sshd only allows publickey-based authentication and the only accepted key types are FIDO2/U2F hardware-backed keys, it's probably more secure again (less attack surface).
With a plain VPN like WireGuard when they get access to your network, they don't have plain ssh, not to mention root ssh access to hosts. This is a serious issue.
Tailscale SSH has caused me other problems in the past because it takes over port 22. I'm not a fan.
It takes over port 22 on the Tailscale interface only. Only had problems with this when I’ve wanted to hit a host’s non-Tailscale ssh service via Tailscale. Otherwise it’s been great for me
Only on the network they create though, right?
I'll stick to my 100% self-hosted Wireguard setup, thank you very much.
Why not tailscale plus head scale for self hosting?
I do not understand this rebuttal.

I also run self-hosted Wireguard. Initially on a Debian box, nowadays it is integrated into my router (admittedly, this is closed source). For around 6 years at this point.

The whole thing could not be easier and simpler. It has never randomly broken on me. It is fast. It is free. There is no middle man, no vendor.

I never understood the popularity of Tailscale, though that is on me. I'm sure it is a great product, I just never tried it, do not seem the target audience.

What confuses me is the often accompanying, sometimes aggressive anti-selfhosting stance in these sorts of threads. I do not see this in other topics, e.g. someone mentioning they run Jellyfin isn't met with "why not Plex?". Where does that come from? We are on HackerNews, not ProductShillNews, aren't we? I guess self hosting Wireguard is too boring to warrant any further discussion? The VPN equivalent of a Toyota Corolla.

I think Tailscale is popular because of how plug and play it is for most people. Although the main reason I use it over self hosting wireguard is the NAT busting it does, which has so far worked flawlessly for me with no setup aside from installing on both devices. There is nothing wrong with self hosting wireguard, but it doesn't actually do the same job as tailscale.
as someone who uses tailscale: exactly this.

i have my homelab only reachable via tailscale and can access everything i would ever want on the go that way. it was a matter of 15 min to get it all working.

Wireguard by itself also doesn't allow for 2FA or expiring keys. Not as relevant for private use, but some orgs need it for compliance. The idea was always that things like that need to be implemented by an application on top of it, so you end up with something like tailscale eventually.
How do I install wire guard on my mom's Apple TV?
it wasn't meant as a rebuttal, I'm genuinely asking. tailscale + headscale was just recommended to me, hence that's what I'm using for self hosting. is wireguard's client roughly equivalent to tailscale's? especially tailscale's always-on nature is very appealing.
My WireGuard uses (either at home or at work) are very much mobile client to single network

Where Tailscale comes into its own is automatic managing of mesh networking (like an “sdwan” solution). The other thing it excels at is firewall busting - if you have a firewall (with or without address translation) which only allows outgoing traffic to be established (with UDP timeouts for session) then Tailscale also works in a similar way to turn/stun.

If I needed that capability then I’d be looking at Headscale. I don’t need it though.

Remember that this is hackernews, not slashdot. Where the community used to be far smaller and the technology far smaller it was quite normal for everyone to understand basic building blocks of ip addresses, use open source software, wear t-shirts threatening to replace people with a small shell script etc.

It’s not the same community, many people here have no real understanding of computer fundamentals, but instead have expertise in specific narrow areas. They also have little interest in things like free software, but do have an interest in building a new billion dollar company to sell to a behemoth.

I also tested tailscale, headscale and netbird and found all these drain your battery on mobile.

Others report no issues but I had massive drain on iOS even with only 4 connections open.

Native wireguard is unnoticeable.

haha self hosted wireguard, an opportunity to find out AllowedIPs: 0.0.0.0/0 does the opposite of what you think it will do
Allow any (ip4) traffic to enter the tunnel, install a route in the default routing table to make that happen (well the second depends on the client)

Does it do the opposite of that?

If you don't care to invest half an hour into learning some basics of how computer networking and in particular CIDR notation and subnet masks work, maybe it is not for you.
i really dislike that there is no way to do dhcp for new clients and that i have to manually define peers in each "exit node"
Because wireguard to tailscale is like git to git GUIs. It's solid base but never should be used separately without a proper wrapper if one wants to keep one's sanity.
tailscale ssh: replacing a 25-year-old battle-tested codebase with a startup's Go rewrite and then acting surprised when it has bugs
(comment deleted)
>>> We would like to thank Anthropic and Ada Logics for reporting this issue.

it seems anthropic also use tailscale or it's just being discovered by the mythos model?

I presume Ada Logics has access to Anthropic's Mythos model via Project Glasswing, and Ada Logics discovered this exploit during their vulnerability research.
pure logic error, the undergoing tailscale rust rewrite can't help this too:)
If it used one of the standard arguments-handling crates (e.g. "clap") there's no way it can happen.
If it used one of the standard system APIs for looking up user accounts (e.g. getpwnam(3)), there's no way it can happen.
Sadly, yet another path to root via Tailscale.

If their scope grows, and they run so much as root, it won't be their last.

> Tailscale SSH now rejects usernames with leading dashes.

Is the proper fix not restricting users not possible in these poorly designed ancient systems?

Similarly re another issue: why not just fix the permission issues instead of restricting users?

> Tailscale now disallows the use of UIDs or numeric-only usernames via SSH to avoid this ambiguity

The proper fix would be to not use getent CLI tool in their logic, but instead use proper system APIs for looking up user account entries, like one of earlier comments here already mentions. This is shocking amateur hour!

My guess is they hastily threw together something hacky in early development, and forgot to replace it with a real, safe solution later.

There is a whole class of security issues where fixes are worse that the issues themselves. Case in point, the OpenSSH itself that sends 100 packets on each keystroke to avoid timing attacks.
> usernames were passed as arguments to getent(1) to retrieve the corresponding passwd entry

Always try to use actual API/system calls (in this case getpwnam) instead of calling sub-processes.